The 3-day rule: how long to let a campaign struggle before you kill it

The single most expensive habit I see in-house teams repeat: killing a new campaign on day two because the numbers look rough.

Ad platforms need a learning phase. Meta’s delivery system alone typically needs 50 conversions per ad set to exit learning and find its footing — cutting the budget or pausing the ad resets that clock to zero. You pay for the learning phase again, from scratch, every time you panic.

My rule of thumb: give a new campaign a minimum of 3 full days and, ideally, one full weekly cycle before judging it — Sunday behavior is not Tuesday behavior, and judging a campaign on a Tuesday-only sample will mislead you either direction.

What I do watch daily, even inside that window:

• Spend pacing — is the platform actually spending the budget, or is it starved by too-narrow targeting?
• CPM trend — a CPM climbing daily with flat performance is an early warning, not a verdict.
• Frequency — if the same small audience is seeing the ad 4+ times in three days, that’s a targeting problem you can fix before day 3, not a reason to kill the whole campaign.

The 3-day rule isn’t patience for its own sake. It’s the minimum time the algorithm needs to do its job — and killing it early is just paying tuition twice.